Predicting Personal Metabolic Responses to Food Using Multi-omics Machine Learning in over 1000 Twins and Singletons from the UK and US: The PREDICT I Study (OR31-01-19)

Abstract

ObjectivesGlycemic, insulinemic and lipemic postprandial responses are multi-factorial and contribute to diabetes, obesity and cardiovascular disease. The aim of the PREDICT I study is to assess the genetic, metabolic, metagenomic, and meal-context contribution to postprandial responses, integrating the metabolic burden and gut microbiome to predict individual responses to food using a machine learning algorithm. MethodsA multi-center dietary intervention study of 1000 individuals from the UK (unrelated, identical and non-identical twins) and 100 unrelated individuals from the US, assessed postprandial (0–6 h) metabolic responses to sequential mixed-nutrient dietary challenges in the clinic. Baseline data included metabolomics, genomics, gut metagenomics and DXA body composition. Glycemic responses to 5 duplicate isocaloric meals of different macronutrient content and self-selected meals (>100,000), were tested at home using a continuous glucose monitor (CGM). Interim analysis of the genetic contributions was performed in 110 identical and 25 non-identical twin pairs. ResultsInter-individual variability in postprandial metabolic responses (glucose, insulin and triacylglycerol (TG)) was high in the clinic setting: incremental area under the curve IQR (median) was; glucose (0–2 h) 2.09 (1.95) mmol/L.h, insulin (0–2 h) 47.0 (67.6) mIU/L.h and TG (0–6 h) 2.34 (2.38) mmol/L.h. The unadjusted genetic contribution for the glucose, insulin and TG responses were 54%, 29% and 27% respectively. A predictive algorithm was developed and interim analyses, using at home CGM data, found that 46% of overall variation in glycemic responses could be predicted from meal content, meal context and individual baseline characteristics excluding genetic and microbiome factors. Only 29% of variation could be explained by the macronutrient content of the meal. ConclusionsThis is the most comprehensive postprandial study performed to date. The large and modifiable variation in metabolic responses to identical meals in healthy people explains why ‘one size fits all’ nutritional guidelines are problematic. By collecting information on glucose responses to >100,000 meals we will have excellent power to use machine learning to optimise and predict individual responses to foods. Funding SourcesNIHR, Wellcome Trust, Zoe Global Ltd.